Unsupervised Domain Adaptation through Self-Supervision

Abstract

This paper addresses unsupervised domain adaptation, the setting where labeled training data is available on a source domain, but the goal is to have good performance on a target domain with only unlabeled data. Like much of previous work, we seek to align the learned representations of the source and target domains while preserving discriminability. The way we accomplish alignment is by learning to perform auxiliary self-supervised task(s) on both domains simultaneously. Each self-supervised task brings the two domains closer together along the direction relevant to that task. Training this jointly with the main task classifier on the source domain is shown to successfully generalize to the unlabeled target domain. The presented objective is straightforward to implement and easy to optimize. We achieve state-of-the-art results on four out of seven standard benchmarks, and competitive results on segmentation adaptation. We also demonstrate that our method composes well with another popular pixel-level adaptation method.

References in corpus (19)

• Rethinking Atrous Convolution for Semantic Image Segmentation
• Deep Visual Domain Adaptation: A Survey
• Progressive Growing of GANs for Improved Quality, Stability, and Variation
• NIPS 2016 Tutorial: Generative Adversarial Networks
• Understanding deep learning requires rethinking generalization
• CyCADA: Cycle-Consistent Adversarial Domain Adaptation
• Domain Separation Networks
• Asymmetric Tri-training for Unsupervised Domain Adaptation
• A DIRT-T Approach to Unsupervised Domain Adaptation
• Unsupervised Cross-Domain Image Generation
• Domain Adaptation for Visual Applications: A Comprehensive Survey
• Many Paths to Equilibrium: GANs Do Not Need to Decrease a Divergence At Every Step
• Geometry of Optimization and Implicit Regularization in Deep Learning
• Stronger generalization bounds for deep nets via a compression approach
• Stabilizing GAN Training with Multiple Random Projections
• What is Local Optimality in Nonconvex-Nonconcave Minimax Optimization?
• Self-ensembling for visual domain adaptation
• Colorization as a Proxy Task for Visual Understanding
• Boosting Supervision with Self-Supervision for Few-shot Learning

Cited by in corpus (57)

• On the Opportunities and Risks of Foundation Models
• Towards Out-Of-Distribution Generalization: A Survey
• Automatic Sleep Staging of EEG Signals: Recent Development, Challenges, and Future Directions
• Self-supervised Knowledge Distillation for Few-shot Learning
• Test-Time Training with Self-Supervision for Generalization under Distribution Shifts
• Just Pick a Sign: Optimizing Deep Multitask Models with Gradient Sign Dropout
• Unsupervised Domain Adversarial Self-Calibration for Electromyographic-based Gesture Recognition
• Self-Supervised Policy Adaptation during Deployment
• A Survey of Deep Learning for Scientific Discovery
• MoCo-CXR: MoCo Pretraining Improves Representation and Transferability of Chest X-ray Models
• Test-Time Adaptation to Distribution Shift by Confidence Maximization and Input Transformation
• Rethinking Distributional Matching Based Domain Adaptation
• Prototypical Pseudo Label Denoising and Target Structure Learning for Domain Adaptive Semantic Segmentation
• Cross-domain Self-supervised Learning for Domain Adaptation with Few Source Labels
• Self-Supervised Learning for Domain Adaptation on Point-Clouds
• Self-Supervised Visual Learning by Variable Playback Speeds Prediction of a Video
• Source Data-absent Unsupervised Domain Adaptation through Hypothesis Transfer and Labeling Transfer
• A Review of Single-Source Deep Unsupervised Visual Domain Adaptation
• Learning from Temporal Spatial Cubism for Cross-Dataset Skeleton-based Action Recognition
• Auxiliary Signal-Guided Knowledge Encoder-Decoder for Medical Report Generation
• Vision-Language Navigation with Self-Supervised Auxiliary Reasoning Tasks
• Anatomy-guided domain adaptation for 3D in-bed human pose estimation
• Prototypical Cross-domain Self-supervised Learning for Few-shot Unsupervised Domain Adaptation
• Learning from Extrinsic and Intrinsic Supervisions for Domain Generalization
• AdaStereo: A Simple and Efficient Approach for Adaptive Stereo Matching
• Open-Set Hypothesis Transfer with Semantic Consistency
• A Learnable Self-supervised Task for Unsupervised Domain Adaptation on Point Clouds
• Delving Deep into the Generalization of Vision Transformers under Distribution Shifts
• EC-GAN: Low-Sample Classification using Semi-Supervised Algorithms and GANs
• From Local Structures to Size Generalization in Graph Neural Networks
• ePointDA: An End-to-End Simulation-to-Real Domain Adaptation Framework for LiDAR Point Cloud Segmentation
• Curriculum Manager for Source Selection in Multi-Source Domain Adaptation
• Learning to Exploit Multiple Vision Modalities by Using Grafted Networks
• Domain Adaptive Semantic Segmentation with Self-Supervised Depth Estimation
• Cross-Domain Sentiment Classification with In-Domain Contrastive Learning
• Graceful Degradation and Related Fields
• Gradual Domain Adaptation in the Wild:When Intermediate Distributions are Absent
• Source-Free Adaptation to Measurement Shift via Bottom-Up Feature Restoration
• Self-Supervised Euphemism Detection and Identification for Content Moderation
• Emotional Semantics-Preserved and Feature-Aligned CycleGAN for Visual Emotion Adaptation
• Unsupervised Multimodal Video-to-Video Translation via Self-Supervised Learning
• One-bit Supervision for Image Classification: Problem, Solution, and Beyond
• What Does Rotation Prediction Tell Us about Classifier Accuracy under Varying Testing Environments?
• Neural-Symbolic Solver for Math Word Problems with Auxiliary Tasks
• Multilevel Knowledge Transfer for Cross-Domain Object Detection
• Multi-source Few-shot Domain Adaptation
• Curriculum CycleGAN for Textual Sentiment Domain Adaptation with Multiple Sources
• An Efficient Method for the Classification of Croplands in Scarce-Label Regions
• RPCL: A Framework for Improving Cross-Domain Detection with Auxiliary Tasks
• Pareto Navigation Gradient Descent: a First-Order Algorithm for Optimization in Pareto Set
• DIRA: Dynamic Domain Incremental Regularised Adaptation
• Online Domain Adaptation for Continuous Cross-Subject Liver Viability Evaluation Based on Irregular Thermal Data
• A Multi-view Perspective of Self-supervised Learning
• Synthetic-to-Real Domain Adaptation for Lane Detection
• Zero-Round Active Learning
• Geometry-Aware Self-Training for Unsupervised Domain Adaptationon Object Point Clouds
• Test-Time Adaptation with Principal Component Analysis